Build a minimal two hidden layer network with step activation that realizes the following decision boundary and specify all the weights and bias. For step activation function, output is $+1$ if total input $>=$ bias T else output is $-1.$

A$.$ Draw the network architecture. What is the minimum number of hidden nodes required at hidden layer $1$ and hidden layer $2?$

B$.$ Specify all the weights and biases. Weights can be only $-1,1$ or $0$ only.

C$.$ Can this decision boundary be realized with one hidden layer? If yes, how many hidden nodes will be required?

please explain in detail how to draw network architecture and proceed with other questions A$,$B$,$C